Recommendation systems challenges at Twitter scale

Ashish Bansal (Twitter)

4:00pm–4:40pm Wednesday, September 11, 2019

Location: 230 A

Implementing AI

Secondary topics: Machine Learning, Text, Language, and Speech

Average rating:

(4.50, 2 ratings)

Download slides (PDF)

Who is this presentation for?

ML engineers, data scientists, engineering managers, and software engineers

Level

Intermediate

Description

Twitter has amazing and unique content generated at an enormous velocity internationally in multiple languages. A constant challenge is how to find the relevant content for users so that they can engage in the conversation. Many traditional recommendation systems approaches don’t work directly at the scale.

Ashish Bansal provides a window into the world of fast-moving recommendations at massive scale and shares practical ideas on why this is such a fun problem to work on.

Prerequisite knowledge

Familiarity with machine learning and software engineering
Familiarity with NLP and recommendation systems like collaborative filtering (useful but not required)

What you'll learn

Understand how to balance scalability with complexity of algorithms and how to build a RecSys with catalog of 100x Amazon's catalog

Ashish Bansal

Twitter

Ashish Bansal is a senior engineering manager, recommendations, leading recommendations teams working on events and trends at Twitter. He focuses on building scalable ML and recommendation systems. Previously, he was a senior director of data science at Capital One where he used AI/ML to generate insights from vast amounts of data and build interesting B2B, B2C, and enterprise products; he cofounded GALE Partners and headed the Machine Learning Group, building AI/ML based marketing automation products. He helped the company grow from 9 to 120 in two years and set up the India office. He has over 19 years of experience in the technology industry, an MBA from the Kellogg School of Management at Northwestern University and a BTech from IIT BHU.